Background
Ovarian cancer is the most common cause of cancer death from gynecologic tumors and is an important global issue requiring further attention [
1]. Each year, more than 225,000 new ovarian cancers are diagnosed, and more than 140,000 deaths caused by ovarian cancer are estimated to occur worldwide [
2]. The introductions of surgical cytoreduction, platinum-taxane-based chemotherapy, and, most recently, intraperitoneal chemotherapy have significantly increased life expectancy with ovarian cancer. Consequently, the median survival for ovarian cancer patients who undergo optimal surgery and adjuvant chemotherapy can be as high as 60 to 110 months [
3].
However, the occurrence of comorbidities, such as cerebrovascular complications, after the cancer event may exacerbate mortality in cancer survivors. In addition, the median survival after stroke in cancer patients is 4.5 months, and treatment has no survival benefit [
4]. In a large autopsy series, up to 14.6% of cancer patients had cerebrovascular complications [
5]. Cancer and ischemic stroke can cause heavy burdens on the economic and healthcare systems and can disrupt the quality of life in survivors. Therefore, it is very important to identify cancer survivors who have a high risk for stroke and to provide them with close surveillance.
In ovarian cancer patients, there are a few studies and reports mentioning the occurrence of stroke [
6,
7]. However, no large-scale study has examined the association of ovarian cancer and ischemic stroke. Hence, it remains unknown whether ovarian cancer patients exhibit an elevated risk of developing ischemic stroke compared with the general population. Therefore, a population-based cohort study using the Taiwan National Health Insurance (NHI) was conducted to examine this issue.
Methods
Data sources
Taiwan’s NHI program is a universal healthcare program that was implemented in 1995. It currently covers more than 99.5% of the entire population [
8]. Data used in this study were obtained from the research database of NHI, which contains encrypted computerized medical claims, inpatient and ambulatory care records, data from the Registry for Catastrophic Illness Patients (RCIP), basic demographic information, and other healthcare data. The accuracy of diagnoses in this database has been validated for several diseases, including ischemic stroke [
9]. Diseases were coded with the cancer International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) diagnosis codes, 2001 edition. Because the Taiwan NHI research database contains encrypted computerized data for research purposes, the ethics committee of Taipei Veterans General Hospital informed us that this study was exempt from full review and that informed consent from each patient was not required.
Study subjects
Two cohorts, an ovarian cancer cohort and a matched control cohort, were enrolled in this study. The ovarian cancer cohort (ICD-9-CM code 183.X) was identified from the RCIP. For a patient to be enrolled in the RCIP, the diagnosis of ovarian cancer had to be confirmed by histology. The diagnosis was subject to periodic review by the Bureau of NHI. Patients with a new ovarian cancer diagnosis made between 1 January 2003 and 31 December 2011 were included in this study. Subjects younger than 20 years old, those who had been diagnosed with any cerebrovascular disease before ovarian cancer diagnosis, and those who were lost to follow-up were excluded. Each individual in the ovarian cancer group was matched on the basis of age, gender, time of enrollment, and comorbidities related to cerebrovascular events with an individual without ovarian cancer in a comparison cohort, which was also selected from the NHI.
Outcome measures
Occurrence of ischemic stroke was identified after the initial diagnosis of ovarian cancer using medical claims data in the NHI. Ischemic stroke identification using data in the Taiwan NHI is highly accurate compared to other insurance databases and is valid for population-based research [
9]. In our study, identification was made on the basis of ischemic stroke coding (ICD-9-CM code 436, 433.X, 434.X, and 437.1X) accompanied by computed tomographic or magnetic resonance images. Diagnoses lacking confirmatory image were not included. Subjects in both cohorts (ovarian cancer and matched) were followed until a diagnosis of ischemic stroke was made, death occurred, 31 December 2011 was reached, or the patient withdrew from the NHI.
Statistical analyses
Microsoft SQL Server 2012 (Microsoft Corporation, Redmond, WA, USA) and SAS 9.2 software (SAS Institute Inc., Cary, NC, USA) were used for data management and analysis. All statistical analyses were performed using SPSS statistical software version 17.0 for Window (SPSS, Inc., Chicago, IL, USA). Statistical significance was defined as a P value less than 0.05.
The demographic and clinical characteristics of ovarian cancer patients and matched cohorts are presented as the total number (n) and proportion (%). Continuous and categorical variables were compared between groups with the Mann-Whitney U test and Pearson’s chi-squared test, as appropriate. To assess the effect of age, study subjects were divided into two subgroups: age <50 years and age ≥50 years. Hazard ratio (HR) and 95% confidence intervals (CI) were calculated with the Cox proportional hazard models to test the association between ovarian cancer and ischemic stroke; the HR was further adjusted for age and comorbidities. The difference in the cumulative incidence of ischemic stroke between the ovarian cancer patient cohort and the matched cohort was calculated with the Kaplan-Meier method and tested with the log-rank test.
Conventional risk factors for stroke, such as diabetes mellitus and hypertension, which are also possible confounding factors, were further assessed for their effects on the association of ovarian cancer and subsequent ischemic stroke in univariate Cox proportion hazards models. Additionally, therapeutic modalities, such as surgery and chemotherapy, were put into Cox models as time-dependent covariates to avoid immortal time bias. All risk factors with P values <0.1 in the univariate model were further entered into the multivariate analysis.
Discussion
This is the first population-based study to determine the risk of new onset ischemic stroke after an ovarian cancer diagnosis. Our data revealed an increased risk of developing ischemic stroke with an adjusted HR of 1.49 among patients with ovarian cancer. Significant risk factors for ischemic stroke included age ≥50 years, diabetes mellitus, hypertension, and chemotherapy treatment.
Up to 40% of stroke cases in cancer are cryptogenic related, possibly induced by the underlying malignancy [
10]. For cancer patients, RRs of developing venous thrombosis of four- to seven-fold have been reported compared with matched controls [
11]. However, cancer is a broad disease and different types of cancer may have different risks. Generally, pancreas, brain, lung, and ovarian cancers have been associated with the higher risks of developing venous thrombosis [
12]. In addition, the presence of disseminated intravascular coagulation in patients with ovarian cancer may indicate a hypercoagulative status [
13]. One direct evidence for the hypercoagulative state is the frequently overexpressed tissue factor in ovarian cancer tissue, which could activate the extrinsic coagulation cascade and cause thrombolic events in ovarian cancer patients [
14].
Although the temporal-causal relationship between ovarian cancer progression and hypercoagulation is unknown, high levels of coagulation factors and associated regulatory proteins have been observed in ovarian cancer patients [
15,
16]. Increased levels of circulating mucinous material produced by ovarian tumors might play a role in intravascular hypercoagulation and hyperviscosity [
17]. Together, these findings suggest that a subgroup of ovarian cancer patients may have distinct coagulopathies, making these patients vulnerable for specific vascular events, including ischemic stroke. These findings may also explain why ovarian cancer survivors in our study, especially younger patients who had no established conventional stroke risk factors, possessed a higher risk of subsequent ischemic stroke.
The most well-known and medically important (that is, treatable) risk factors for ischemic stroke include hypertension, diabetes mellitus, dyslipidemia, smoking, and obesity [
18]. Although lifestyle and obesity were not included in the present analysis, some epidemiological studies have demonstrated that hypertension and diabetes are significant risk factors for stroke in several cancers [
19,
20]. Our multivariate analyses revealed a similar result, with additional significant risk factors, including age ≥50 years old and chemotherapy treatment. These results suggest that conventional stroke risk factors remain important in the pathogenesis of stroke in ovarian cancer patients.
Our study showed that ovarian cancer patients receiving chemotherapy, particularly platinum-based regimens, might have an additionally increased stroke risk. This effect was insignificant in non-platinum-based regimens. In prior studies, chemotherapy-associated increased stroke was observed in patients with head-and-neck cancer [
19,
21], breast cancer [
22], and urothelial carcinoma [
23]. While the mechanism is still uncertain, some suggested pathophysiologies are associated with increased fibrinopeptide A and decreased fibrinolytic activity [
24], elevated plasma von Willebrand factor [
25], hypomagnesium-associated vascular spasm [
26,
27], endothelial injury [
28,
29], and mononuclear cell-mediated platelet activation [
30]. Moreover, Li
et al. demonstrated that platinum in a chemotherapy regimen, which is also commonly used for treating ovarian cancer patients, may increase the risk of ischemic stroke among cancer patients [
31]. Cerebral infarction after cisplatin-based chemotherapy in ovarian cancer patients has also been reported previously [
32,
33]. A meta-analysis of 38 randomized phase II and III trials showed that cancer patients who received cisplatin-based chemotherapy demonstrated a dose-dependently increased risk (RR 1.67, 95% CI, 1.25 to 2.23;
P = 0.01) of thromboembolism compared with patients who received a non-cisplatin-based regimen [
34]. More research is needed to clarify the role of chemotherapy in the relationship of ovarian cancer and subsequent ischemic stroke, as well as to determine whether it is necessary to use a prophylactic antiplatelet agent in high-risk patients during the chemotherapy period.
Our study revealed that ovarian cancer patients appeared to have a higher risk of stroke soon after cancer diagnosis, and the increased risk persisted throughout the follow-up period. A similar phenomenon was also observed in survivors of head-and-neck cancer [
19,
21], cervical cancer [
35], breast cancer [
36], and Hodgkin lymphoma [
37]. The stroke risk imposed by ovarian cancer seemed to be more prominent in people of younger age, with adjusted HRs of 2.28 and 1.33 in the age groups of <50 years and ≥50 years, respectively. These findings suggest that, with increasing age, the age factor becomes more important than cancer as a risk factor of stroke.
This study has several limitations. First, lifestyle variables and behavioral factors, such as tobacco and alcohol use, body mass index, dietary habits, and biochemistry profiles including serum D-dimer level and disseminated intravascular coagulation profiles, were not available in the claims data of the Taiwan NHI. Second, cancer staging and histology were not available in this database, and their association with ischemic stroke could not be identified. Finally, information on the cause of death was not available; consequently, we cannot determine the impact of stroke on cancer-related mortality. Despite these limitations, our study was based on a nationwide, population-based database that could identify all cases of ovarian cancer and ischemic stroke in the study period. The large sample size in our study contributed to substantial statistical power and revealed a clear association between ovarian cancer and ischemic stroke, with subtle statistically significant differences between the two cohorts.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
ASK and CJL had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. ASK, CJT, and CJL designed the study. CMY and CJL acquired the data and performed statistical analysis. ASK, CJT, HHW, TJC, and CHT gave the final interpretation of the results. ASK and CJT drafted the manuscript. CJT, HHW, VYS, YTC, SHC, and LYH gave critical revision of the manuscript for important intellectual content. CMY, TJC and CHT gave administrative, technical, and material support. TJC, CHT and CJL were the study supervisors. CJL acts as guarantor and accepts responsibility for the integrity of the work as a whole. All authors read and approved the final manuscript.